A lexical comparison of South African sign language and potential lexifier languages

(1)

by

Andries van Niekerk

Thesis presented in partial fulfilment of the requirements for the degree Masters in General Linguistics at the University of Stellenbosch

(2)

DECLARATION

By submitting this thesis electronically, I declare that the entirety of the work contained therein is my own, original work, that I am the sole author thereof (save to the extent explicitly otherwise stated), that reproduction and publication thereof by Stellenbosch University will not infringe any third party rights and that I have not previously in its entirety or in part submitted it for obtaining any qualification.

Andries van Niekerk

March 2020

(3)

ABSTRACT

South Africa’s history of segregation was a large contributing factor for lexical variation in South African Sign Language (SASL) to come about. Foreign sign languages certainly had a presence in the history of deaf education; however, the degree of influence foreign sign languages has on SASL today is what this study has aimed to determine. There have been very limited studies on the presence of loan signs in SASL and none have included extensive variation. This study investigates signs from 20 different schools for the deaf and compares them with signs from six other sign languages and the Paget Gorman Sign System (PGSS).

A list of lemmas was created that included the commonly used list of lemmas from Woodward (2003). The signs were elicited from informants and documented based on their phonological properties: handshape, movement, location, and orientation. SASL lexical variety was documented. Sign types were identified by comparing the different signs for a lemma with each other; signs that differed in more than one phonological parameter were classified as different sign types. These sign types were then compared with counterparts in six potential lexifier sign languages, American Sign Language (ASL), British Sign Language (BSL), Irish Sign Language (ISL), German Sign Language (DGS), Flemish Sign Language (VGT), and Dutch Sign Language (NGT) and PGSS. Signs that are heavily influenced by iconic devices were removed from the final comparison. Loan signs were identified on the basis of phonological similarity.

The results showed the highest percentage of borrowings from BSL at 15.9%, followed by ASL with 12.6% and VGT at 11.7%. The results indicated that 65.4% of the sign types were influenced to some degree by foreign sign languages or PGSS. There is a substratum of signs that did not match with any of the potential lexifier languages or PGSS and their origins are uncertain; they possibly emerged naturally or were borrowed from a language that was not included in this study.

(4)

OPSOMMING

Die geskiedenis van segregasie in Suid-Afrika was 'n groot bydraende faktor tot die leksikale variasie in Suid-Afrikaanse Gebaretaal (SAGT). Buitelandse gebaretale het beslis 'n teenwoordigheid gehad deur die verloop van dowe onderwys; die invloed wat vreemde gebaretale op SAGT vandag het, is egter waarop hierdie studie fokus. Daar is beperkte studies oor leen-gebare in SAGT, en nie een van hulle het die uitgebreide variasie inherent in SAGT ingesluit nie. Hierdie studie ondersoek gebare van 20 verskillende skole vir dowes en vergelyk dit met gebare uit ses ander gebaretale asook die Paget Gorman Sign System (PGSS).

'n Lys van lemmas is opgestel wat die algemeen gebruikte lys uit Woodward (2003) bevat. Die gebare is vanuit informante ontlok en gedokumenteer op grond van die volgende fonologiese eienskappe: handvorm, beweging, ligging en oriëntasie. SAGT leksikale variasie is gedokumenteer. Gebare tipes is geïdentifiseer deur die verskillende gebare vir 'n lemma met mekaar te vergelyk; gebare wat met meer as een fonologiese parameter verskil het, is as verskillende gebare klassifiseer. Hierdie gebare tipes is vervolgens vergelyk met eweknieë in ses gebaretale, Amerikaanse Gebaretaal (ASL), Britse Gebaretaal (BSL), Ierse Gebaretaal (ISL), Duitse Gebaretaal (DGS), Vlaamse Gebaretaal (VGT), en Nederlandse Gebaretaal (NGT) en PGSS. Gebare wat sterk beïnvloed is deur ikoniese stelsels, is uit die finale vergelyking verwyder. Leen-gebare is op grond van fonologiese ooreenkomstes geïdentifiseer. Die resultate toon dat die hoogste persentasie lenings vanaf BSL is met 15,9%, gevolg deur ASL met 12,6% en VGT met 11,7%. Die resultate het aangedui dat 65,4% van die gebare wel tot 'n mate beïnvloed is deur vreemde gebaretale of PGSS. Daar is 'n substraat van gebare wat nie ooreenstem met enige van die moontlike gebaretale of PGSS nie, en die oorsprong daarvan is onseker; hulle het moontlik natuurlik na vore gekom of is geleen uit 'n taal wat nie by hierdie studie ingesluit is nie.

(5)

ACKNOWLEDGEMENTS

I would like to thank the National Institute for the Deaf for their permission to use the data from their dictionary project. Additionally, I wish to thank the students for their willing participation in the dictionary project’s data collection process.

I certainly would not have been able to complete this without the extremely knowledgeable guidance from Anne Baker and Kate Huddlestone. Anne’s insights on clarity and her incredible knowledge enabled me to look at this thesis from outside my bubble and with Kate’s guidance I could create the proper framework within which a thesis could be built.

The support I received from colleagues were invaluable and my thanks go to Ilse Bloem, Lynette Victor, Van Zyl Naude, Dirkie Ebersohn, and Simon Ndaba. As many researchers in sign language are aware, videography is a key part; my thanks go to Jaco Engelbrecht for his exceptional standard of video recording and editing.

Finally, I would like to thank my family for their patience and encourage them to view the past two years as practice for the next four years.

(6)

Chapter 1 Introduction

1.1. Background

The National Institute for the Deaf (NID) in the Western Cape, South Africa has a department that provides post-school training to Deaf people. As part of their orientation programme, students from different schools for the deaf from different provinces across South Africa are asked to stand up on the stage and show the signs for a few basic words as used at their school. The students regularly express amazement at the variety of signs that emerge. The lexical variation within South African Sign Language (SASL) is in fact apparent to any observer.

SASL is the language of the South African Deaf community. The number of users vary depending on the source consulted; 243 000 (SA Census 2011), 500 000 (DeafSA in Magongwa, 2010), 1 000 0001_{(DeafSA). SASL has been recognised by the 1996 constitution} of South Africa as a minority language and further by the South African Schools Act of 1996 as a language of teaching and learning. Apart from these policies, very little has been done in terms of language planning until SASL became recognised as a school subject, with nationwide SASL curriculum rollout in 2015 (see Morgan, Glazer & Magongwa 2016). SASL learning materials are currently distributed to all the schools for the deaf by the Department of Basic Education; all schools will have materials utilising the same lexicon and this will undoubtedly have an effect on the variation found in the language.

It is an interesting question where the current lexical variation in SASL may come from. The variation appears to be linked to the history of SASL and that in turn is closely linked to the history of deaf education in South Africa which has been influenced by several different European countries.

The Irish Dominican Order established the first school for the deaf in South Africa in 1863. This school was a signing school (Aarons and Akach, 1998) until 1880 when an international conference was held in Milan, Italy. At this conference it was argued that sign languages were not equal to spoken languages and that deaf children would be disadvantaged if they used a sign language. This conference proposed that only spoken languages be used in deaf education; sign languages were then forbidden.

South Africa is one of the countries that adopted the resolutions of this conference and sign languages were then not allowed in schools for the deaf. This was enforced in schools with

(9)

white deaf children, but not very strictly in schools with other races2_{(Aarons and Akach, 1998).} This provided ample opportunity for lexical variation which was further exacerbated by the government policy to assign specific racial groups to specific schools (Van Herreweghe and Vermeerbergen, 2010). Despite SASL being forbidden at school it was still being used between deaf people (Aarons and Akach, 1998) and more so at schools that were less strictly monitored. However, there was limited contact between the different schools and thus the language variants at the various schools for the deaf developed largely independently from other schools for the deaf, more so between schools with children of different races. There was very little opportunity for SASL users to be exposed to other variants.

Schools for the deaf were primarily established and run by religious organisations (Aarons and Akach, 1998). Parents who were averse to their children attending the Irish Dominican Catholic school approached the Dutch Reformed Church who established the second school for the deaf in 1881. Throughout the 20th century, many more schools were established, the clear majority by religious orders: Dutch Reformed Church, Irish Dominican, German Dominican. Given the history of SASL and deaf education, it is reasonable to assume that there would be large variation in, at least, the lexicon. Schools for black children used the Paget Gorman Sign System, a sign system invented to manually represent English, while other types of communication were introduced in other schools. Signed Exact English (SEE) was introduced into the VN Naik school for the deaf. SEE is a form of simultaneous communication where English is the primary language; the English word order is used. Signs are used at the same time as the English is spoken and all of the signs are borrowed from American Sign Language (Van Herreweghe and Vermeerbergen, 2010).

There are several factors that may have contributed to the lexical variation in SASL. It is likely that the lack of contact between schools for the deaf was the primary factor that contributed to the variation. A second factor is that schools for the deaf sent staff to other countries to learn about deaf education or invited experts and likely appropriated the country’s sign language resources. These teachers and administrators likely had little knowledge of SASL

(10)

introduced their own signs into the varieties used at the schools they were responsible for. It is regrettable that the history of the schools has not been well documented.

The Wikipedia entry for SASL3 mentions that it is a part of the BANZL language family i.e. British Sign Language, Australian Sign Language, and New Zealand Sign Language; this claim is unsubstantiated. There is nothing known about the history of signing in South Africa before the nineteenth century. There is also not much known about the current influences on SASL; one can make assumptions based on the history of the schools. The Dutch Reformed Church established many schools for the deaf and one may assume that Dutch Sign Language had a large influence. British Sign Language and American Sign Language were not so prominent in the history of schools for the deaf and one might assume less of an impact from these two sign languages. The Dominican orders provide grounds for possible influences from Irish Sign Language, and German Sign Language. This study aims to provide a clearer understanding of the influence that other sign languages have had on SASL.

1.2. Aim of the research

This thesis will investigate lexical borrowing, as related to lexical variation, in SASL. It aims to determine the extent of lexical borrowing by comparing signs used at identified schools for the deaf in different provinces of South Africa with signs from other sign languages that were deemed likely to have influenced SASL. The identified sign languages are American Sign Language (ASL), British Sign Language (BSL), Sign Language of The Netherlands (NGT), Flemish Sign Language (VGT), German Sign Language (DGS), Irish Sign Language (ISL), and a sign system known as the Paget Gorman Sign System (PGSS).

The signs will be phonologically described and compared with counterparts from other sign languages; similarities and differences will be documented, and the results will indicate the extent of lexical overlap between SASL and the other sign languages.

1.3. Outline of the thesis

In chapter two a review will be done of the literature on variation in sign languages and sign language lexical comparisons. This review will result in a presentation of the research questions that the study will seek to answer. Chapter three will lay out the methodology and analysis procedure and outline the NID dictionary project which provides the data for this study, elaborating on the informants and data collection procedures followed in the project.

3_{https://en.wikipedia.org/wiki/South_African_Sign_Language}_{- this is included to illustrate the misinformation}

(11)

The core of the methodology revolves around the phonological parameters of signs and these parameters will be described. The results will be presented in chapter four and will be discussed in chapter five.

(12)

Chapter 2 Literature Review

This chapter will review other research that has examined variation within a sign language and research that has involved a comparison of the lexicons of languages, focussing on sign languages. In 2.1 the variation in sign languages is discussed. Secondly, in 2.2, the field of comparative lexicostatistics will be discussed. In 2.3 the comparison of sign language lexicons will be detailed and in 2.3.1 the studies of one of the pioneers of this field in sign languages, James Woodward, will be reviewed. Important work has been done with Australasian sign languages and section 2.3.2 is devoted to this. The studies on other sign languages are discussed in 2.3.3 and followed, finally, by a review of studies on South African Sign Language in 2.3.4, and a brief conclusion in 2.4. The chapter ends with a section (2.5) detailing the research questions that the study seeks to answer.

2.1. Variation in sign languages

Variation is predominantly observable in the lexicons for sign languages, although morphological and syntactical variation also occur.

The history of SASL (see chapter 1) with the segregated schools for the deaf, lack of contact between the schools, and apparent influences from foreign sign languages, provides the background for variation to exist. The establishing organisations possibly introduced their own learning material and sign languages to the schools they established. A second, and arguably more prominent, factor in the variation is the schools for the deaf themselves. Sutton-Spence and Woll (2010) coined the term schoolization to explain the changes a sign language undergoes as it is primarily transmitted through interaction with peers at a school for the deaf. This can be seen in NGT where regional variants of this sign language developed at schools for the deaf in the Netherlands (Schermer, 2004). It is also a logical assumption that South Africa’s 43 schools for the deaf, which includes units for deaf children at schools for hearing children, across an extensive geographical area has resulted in considerable variation.

Penn and Reagan (1994) discuss lexical variation within SASL, in a report which was the culmination of a seven-year project to document sign varieties of SASL. Data for 2500 lexical items were collected from deaf adults in eleven Deaf communities across South Africa. The researchers found on average six variants per word. This variation raised questions about the number of sign languages in South Africa. Aarons and Akach (1998) disputed notions of

(13)

multiple sign languages in South Africa ascribing variation to the history of deaf education in South Africa.

Standardisation is a process which can lead to a decrease in variation. It can occur in a direct or indirect manner. Signers who frequently have contact outside of their locale may readily adopt new signs while others who are more isolated may not (Sutton-Spence and Woll, 2010); this is an example of natural standardisation. The contribution of the internet allowing exposure to other SASL variants and the transfer of students between schools further allows for natural standardisation. In contrast to this, authorities may direct standardisation by prescribing dictionaries and learning material, and this is generally highly controversial (Baker, van den Bogaerde, Pfau, Schermer, 2016).

The Netherlands underwent a process of directed standardisation despite objections from sign language researchers and the deaf community. The deaf community experienced standardisation as a system where correct signs were identified and promoted as the standard and other variants deemed incorrect. This was not the intention of the government and even less so of the working group that was established to drive the process. This illustrates the pitfalls of directed standardisation.

What is the current status of SASL in terms of standardisation? There is no official process of standardisation, yet directed standardisation is to a certain extent indeed occurring. SASL has become a school subject (Steyn, 2015) and the learning material used is approved by a central committee. This committee is effectively contributing to the standardisation of SASL by approving specific materials to be used at schools for the deaf.

The lexical variation in SASL creates a conundrum for lexicographic research; which sign to select as representative of a lemma which is the dictionary form or citation form of a word. The present study does not select a single sign to represent a lemma, but rather uses all the available signs in the data and compares them individually with the potential lexifier languages.

(14)

are core to the human experience; these lists include lemmas like tooth and sky (Shosted, 2000) and therefore likely to be part of the lexicon. This method compares core, stable vocabularies to establish cognates, that is word pairs from different languages that may have similar linguistic roots. By examining cognates, it is possible to establish the relationship between the two languages or varieties. Cognates are word pairs from different languages that may have similar linguistic roots.

The percentage of similarity gives an indication of the relatedness of the languages being compared (Gudschinsky, 1956). Previous researchers have determined that dialects of the same language should have an 81% to 100% rate of cognates, 36% to 81% indicate that the languages are from the same language family, and less than 36% means that the languages are unrelated (Crowley, 1992). There are criticisms of this method, most notably by Dixon (1997) who argues that comparing core vocabularies, which is what the Swadesh list claims to be, is no guarantee of genetic relationships.

This study will not be attempting to identify cognates. Finding cognates is a different process and includes more than merely performing phonological comparisons (Parkhurst and Parkhurst, 2003) while this study focuses chiefly on phonological features. This study does not have sufficient historical data to reliably determine cognates which is a limitation as cognates is a more accurate representation of historical language contact. However, the methodology used in cognate studies is very relevant to this study.

2.3. Comparative lexicostatistics in sign languages

The original Swadesh wordlist is not immediately applicable for comparisons of sign language lexicons. Sign languages are different from spoken languages in the medium of their production and transfer. There are different lexemes in a spoken language, articulated by audible phonemes. Iconicity is quite low in spoken languages. Iconicity is the degree to which a sign is conformant with the real-world object it symbolizes (Baker et al., 2016). If someone with no knowledge of SASL was asked to produce the sign for sleep, they could be expected to produce it accurately. It is not necessary to know SASL to be aware of the gesture that is prevalent in society for the concept sleep.

Sign languages utilise the arms and hands as visual phonemes (called cheremes by Stokoe and Kuschel, 1979) and utterances are produced in the space in front of or in contact with the body. Whereas in spoken languages the names of different body parts e.g., ear, arm and so on, are not interpretable to anyone unfamiliar with the language sign languages are different in this

(15)

respect. The vast majority of sign languages refer to different body parts by pointing at the relevant body part with the index finger. A person totally unfamiliar with the sign language would very likely be able to decipher the meaning.

Figure 2.1 EAR in Syrian Sign Language (left) and Russian Sign Language (right)4

Thus, the sign EAR5 in two sign languages is very likely to be identical without there being necessarily any relationship between these languages (see Figure 2.1). To address this issue with the Swadesh list, Woodward (1978) modified the list (see Table 2.1) in order to compare American Sign Language (ASL) and French Sign Language (LSF). Comparing these two sign languages without excluding the indexical signs would result in false positives. This list from Woodward has been a basis in comparative lexicostatistic research for sign languages.

(16)

1. all 26. grass 51. other 76. warm

2. animal 27. green 52. person 77. water

3. bad 28. heavy 53. play 78. wet

4. because 29. how 54. rain 79. what

5. bird 30. hunt 55. red 80. when

6. black 31. husband 56. right 81. where

7. blood 32. ice 57. river 82. white

8. child 33. if 58. rope 83. who

9. count 34. kill 59. salt 84. wide

10. day 35. laugh 60. sea 85. wife

11. die 36. leaf 61. sharp 86. wind

12. dirty 37. lie 62. short 87. with

13. dog 38. live 63. sing 88. woman

14. dry 39. long 64. sit 89. wood

15. dull 40. louse 65. smooth 90. worm

16. dust 41. man 66. snake 91. year

17. earth 42. meat 67. snow 92. yellow

18. egg 43. mother 68. stand 93. full

19. grease 44. mountain 69. star 94. moon

20. father 45. name 70. stone 95. brother

21. feather 46. narrow 71. sun 96. cat

22. fire 47. new 72. tail 97. dance

23. fish 48. night 73. thin 98. pig

24. flower 49. not 74. tree 99. sister

25. good 50. old 75. vomit 100. work

Table 2.1 Woodward (2003) wordlist for research with sign languages

2.3.1. Early sign language comparative studies

Woodward’s (1976) pioneering comparison of American Sign Language (ASL) and French Sign Language (LSF) was not based on the Swadesh list. Woodward aimed to determine the historical relatedness between ASL and LSF. There was plausible basis for there being a relationship between these two languages due to French deaf education having influenced deaf education in America. It would therefore be unlikely for similar signs between these sign languages to be due to chance. Woodward used data from an LSF dictionary containing 873 images. However, Woodward did not have ASL data for all of these dictionary entries. The purpose of the research was to identify similar signs across the two languages and, additionally, to determine how the sign may have changed over time.

Woodward compared signs based on the four phonological parameters: handshape, location, palm orientation, and movement; Woodward does not provide reasons for exluding non-manual markers. Each pair of signs that exhibited similarity was described and the differences documented. Initialised signs are produced with a handshape from the manual

(17)

alphabet that corresponds to the written translation of the word. Woodward listed, but did not discuss changes in initialised signs because he wanted to exclude the influence of other languages. This study (Woodward, 1976) was not strictly a comparison, but rather an investigation into the history of ASL and found 54 signs that were related through a process of historical change with LSF tending to have the older form.

In later work Woodward did use an adaptation of the Swadesh list. For example, sign language varieties in Costa Rica were investigated in Woodward (1991). Signs were collected from deaf signers in various geographical areas of Costa Rica. The modified Swadesh list (Woodward, 1978) was the only data used in the paper. It is relevant to highlight that Woodward could not elicit signs for all of the lemmas. This research identified four different sign language varieties in Cost Rica based on Gudschinsky’s (1956) stipulations (see section 2.2). The four sign language varieties had similar signs for rain, wide, wind, and vomit amongst others. Woodward makes specific mention of the possibility of the influence of iconicity. Woodward did not describe the methodology used.

The relationship between sign language varieties in India, Nepal, and Pakistan was studied by Woodward (1993). The list of 100 lemmas used in this study was the same as the Woodward (1991) study. Sign language varieties of Indian Sign Language examined were those based in New Delhi, Bangalore, Mumbai (formerly known as Bombay), and Calcutta; for Pakistani Sign Language the variety examined was that based in Karachi; for Nepalese Sign Language the variety was based in Kathmandu. Woodward used previously created dictionaries for all of these varieties; the data only had lexical items for 62 lemmas on the list. Printed dictionaries have inherent challenges when used in sign language research. The movement parameter is complex, and it is unlikely that a person unfamiliar with a sign would be able to accurately reproduce the movement of a sign from written instructions. Woodward makes no mention of challenges faced in making the comparison.

(18)

language varieties to represent Indian Sign Language in the data presented a more accurate result. A language comparison that selects a specific language variety to represent a particular language is essentially claiming that the other variants are different languages. Language comparisons should therefore aim to include as many variants as possible in order to have a more complete representation of the language as a whole.

Thai Sign Language (TSL) and its relationship to ASL was investigated by Woodward (1996). Similar to the previous studies mentioned, the modified Swadesh list of lemmas was used. Data was sourced from published materials. For this study, Woodward referred to a standard variety of TSL as Modern Standard Thai Sign Language (MSTSL); this variety was compared with ASL and a 57% rate of cognates was found indicating that the two languages belonged to the same language family (Gudchinsky, 1956). In this study, Woodward verified the data from the dictionaries by recording four signers, the hearing status of these informants were not indicated. The possibility of false cognates was identified which may be due to iconicity. To address this, older signers were interviewed who used a variety prior to the arrival of ASL in Thailand. With these older signers there was a cognate rate of 10% of the 100 lemmas; these ten signs were the same in ASL. These ten lemmas were removed from the comparison and were categorised as being highly iconic; one could not reliably state that the signs were identical due to language contact because the older signers used a variant that existed prior to the arrival of ASL. The previous rate of 57% rate was reduced to 52% after removing the lemmas that resulted in false cognates, which results in a more accurate evaluation of the influence of ASL on MTSL.

The studies by Woodward (1978, 1991, 1993, 1996) were methodologically similar in nature. All the studies used the wordlist that was adapted from the Swadesh list of 200 lemmas. However, Woodward did not consistently note the hearing status of informants and rarely provided information on whether the signers were native signers or had acquired a sign language later. The data is meant to be representative of a language which is why the hearing status and signing level of informants are important. None of the studies discussed went into any detail on how the comparison between lexical items was conducted. Woodward did include sign language varieties of a national sign language in the comparisons, most notably the four Indian Sign Language varieties (Woodward, 1993).

(19)

2.3.2. Comparisons of Australian, New Zealand, and British Sign Languages

McKee and Kennedy (2000) compared signs from ASL, British Sign Language (BSL), Australian Sign Language (Auslan), and New Zealand Sign Language (NZSL). The list of lemmas in Table 2.2 was used and was heavily influenced by Woodward’s modified Swadesh list (Woodward, 1978).

1. all 26. grass 51. other 76. warm

2. animal 27. green 52. person 77. water

3. bad 28. heavy 53. play 78. wet

4. because 29. how 54. rain 79. what

5. bird 30. look for 55. red 80. when

6. black 31. husband 56. correct 81. where

7. blood 32. ice 57. river 82. white

8. child 33. if 58. bug 83. who

9. count 34. kill 59. salt 84. wide

10. day 35. laugh 60. sea 85. wife

11. die 36. leaf 61. sharp 86. wind

12. dirty 37. lie 62. short 87. with

13. dog 38. live 63. sing 88. woman

14. dry 39. long 64. sit 89. wood

15. boring 40. string 65. smooth 90. worm

16. dust 41. man 66. snake 91. year

17. earth 42. meat 67. snow 92. yellow

18. egg 43. mother 68. stand 93. full

19. grease 44. mountain 69. star 94. moon

20. father 45. name 70. stone 95. brother

21. feather 46. narrow 71. sun 96. cat

22. fire 47. new 72. tail 97. dance

23. fish 48. night 73. thin 98. pig

24. flower 49. not 74. tree 99. sister

25. good 50. old 75. vomit 100. work

Table 2.2 McKee and Kennedy (2000) wordlist used to compare ASL, BSL, Auslan and NZSL

Comparing this list to the list used by Woodward (1978) there are relatively few differences: the lemmas rope, louse and dull, were changed to string, bug and boring; hunt

(20)

parameters differed, the signs were said to be completely different. There was an ‘other’ category added for signs that were different, but not in any of the four parameters, e.g., two signs were phonologically identical, but one was produced with one hand and the other with two hands.

McKee and Kennedy found that, if ASL was included in the comparison, there were only 19 signs that were identical across the four languages. Excluding ASL drastically increased this to 64 identical signs. NZSL and Auslan had 77 identical signs; NZSL and BSL had 69 identical signs. The percentage of signs that were cognate, including both identical and related signs, were 87% for NZSL and Auslan and 79% for NZSL and BSL. McKee and Kennedy found a high degree of similarity when comparing signs using the modified Swadesh list.

However, they echoed a concern which had been raised by Dixon (1997), namely that a comparison using only data from the modified Swadesh list may result in a higher incidence of similarity. In order to more accurately compare the lexicon of NZSL with those of ASL, Auslan, and BSL, 200 random everyday lemmas were included in the comparison. Including the random lemmas reduced the rate of cognates for NZSL and Auslan from 87% to 65.5% and for NZSL and BSL from 79% to 62.5%. This may be attributed to the Swadesh list containing lemmas that are seen to be stable (Starostin, 2000). According to Starostin (2000) these stable lemmas tend to be resistant to change and comparisons using these lemmas would tend to yield more similarities. McKee at al. concluded that Auslan, BSL and NZSL were separate languages, but closely related.

Johnston (2003) does not agree with the findings of McKee and Kennedy (2000). Johnston made the same comparison of signs from Auslan, BSL and NZSL but included all variations of a lexeme. Johnston approached the comparison differently and claimed that McKee and Kennedy (2000) compared signs as they were listed in the dictionary and as per the English gloss i.e. the meaning of the sign in English; they did not consider other variants of the sign or other relevant signs with different glosses. Johnston’s approach disregarded the gloss if two signs were the same but had different glosses. Additionally, Johnston identified specific handshapes that were commonly interchangeable and disregarded differences with these specific handshapes, i.e. signs that differed only in one of these handshapes were counted as identical.

Johnston compared the languages using the modified Swadesh list and then 100, 200, and 300 randomly selected entries in the NZSL dictionary. Johnston also did not include

(21)

handedness in his comparison, i.e. a sign that is signed with one hand, mirrored on two hands, or two-handed with one being a weak hand. Johnston illustrated the importance of identifying meaningful phonological variation, which is not always where one phonological parameter is different. He elaborates on this by explaining how signs can be modified for aspect or number, but that this does not modify the base sign. For example, a sign that is modified to indicate plurality should not be used to represent a lemma, but the base sign should be used.

Similar to McKee and Kennedy (2000), Johnston found significant lexical overlap between the languages when using the modified Swadesh list; more than 90% across the three languages. Results for the randomly selected words were also significantly higher than McKee and Kennedy. Johnston states categorically that BSL, Auslan, and NZSL are not separate languages, but rather sister languages with BANZSL (British, Australian and New Zealand Sign Language) as the ‘mother’ sign language.

The reason for the differences between Johnston (2003) and McKee and Kennedy (2000) may be attributed to several factors. Johnston did not limit the comparison to gloss entries and had a wider definition of what would constitute identical or similar signs. Johnston gave no weight to certain attributes of a sign, e.g. whether a sign had a base hand, and points out the variation in a sign’s production between native signers. However, neither Johnston nor McKee and Kennedy made mention of the possibility of false cognates.

2.3.3. Comparisons of the lexicons of other sign languages

Al-fityani and Padden (2008) compared signs from ASL and five sign languages in the Middle East: Jordanian Sign Language (LIU) was compared with Al-Sayyid Bedouin Sign Language (ABSL), Kuwaiti Sign Language (KSL), Libyan Sign Language (LSL), and Palestinian Sign Language (PSL). Signs from published dictionaries for these languages were used with the exception of one, ABSL, where signs were elicited from a native signer. Al-fityani and Padden do not describe the detail of their phonological comparison process. They

(22)

two unrelated sign languages, such as LIU and ASL, than would be the case for two unrelated spoken languages. The two unrelated sign languages they argued were more similar due to the abundance of iconic motivations in sign languages.

Miyamoto and Mori (2015) compared signs from Kenyan Sign Language (KSL) and ASL. They used a dictionary of 2,894 signs but did not rely only on the dictionary entries. All 2,894 lemmas were used to elicit signs from six native signers again. If the newly elicited sign was different from the dictionary, both signs were completely removed from the data. The native signers participated in a workshop prior to the elicitation which taught them about linguistic research principles and in which it was specifically mentioned that influences from other sign languages were to be avoided. It is not clear if signs that are used in KSL but are borrowed from another sign language other than ASL, were excluded in this research purely on that basis, despite being part of the KSL lexicon. In their final comparison only 100 lemmas were used. Miamoto and Mori found that ASL and KSL had between 13% and 22% cognates (see Table 2.3) although this is only for 100 lemmas and not the entire list of lemmas; it is not clear why the comparison was not done with the entire data set.

(23)

Table 2.3 Cognates between KSL and ASL (Miyamoto and Mori 2015:21)

Su and Tai (2009) followed an approach that investigated the role of iconicity in comparative studies for the purpose of establishing historical relationships. In their study they compared Taiwanese, Chinese, Japanese, and American Sign Languages. Similar to the current study, the researchers presented results that included signs with iconic motivations as well as with these signs excluded.

(24)

Figure 2.2 An iconic sign SCISSOR with the relevant handshape in Taiwanese Sign Language (Su and Tai 2009:153)

Su and Tai found 53% of signs to be identical between Taiwan and Japanese sign languages when iconic signs were included; this reduced to 44% when iconic signs were excluded. As expected, there was a reduction in identical signs for all sign languages compared. Woodward’s modified Swadesh list was used, which they claim is insufficient to establish language relationships, but which can partly be addressed by increasing the sample size. However, they also assert that the majority of signs have iconic motivations to different degrees. The presence of iconic motivations does not automatically exclude the possibility of cognates and researchers need to pay attention to the historical relationship between languages.

Parkhurst and Parkhurst (2003) highlights the complexity of determining cognates. It is not merely a matter of determining lexical similarity, but of historical relatedness. Two languages that are related will share more than just lexical items; relatedness will be evident in the morphology, syntax, etc.

Comparing the signs in A and C in Figure 2.3 would result in characterising them as different signs because two phonological parameters are different, the handshape and palm orientation. It is possible to deny that these two signs are cognates if the decision is based solely on the phonological comparison. However, if one is to take into account historical factors, assuming that the variety represented in C is a later variety than that represented in A, mediated by the variety represented in B in Figure 2.3 one can clearly see the link between A and C.

Figure 2.3 MEAT in three dialects of Spanish Sign Language (Parkhurst and Parkhurst 2003:2)

(25)

Parkhurst and Parkhurst compared signs from Spain, Northern Ireland, Finland, and Bulgaria. Their methodology allowed for signs to be cognates, even if they were not similar and therefore, they found a higher rate of cognates than of similarity because they took historical interactions between languages into account. The authors considered historical data and considered the phonological changes signs underwent and if a plausible link could be established, the signs were then marked as cognates. Signs were labelled as “identical or very similar”, “somewhat similar”, or “quite different”; a similar scale to previous research (McKee and Kennedy, 2000; Johnston, 2003). However, it is not made clear what “very similar” or “somewhat similar” would mean. The researchers found a higher incidence of cognates than similar signs for all sign languages compared. Signs with iconic motivations were removed from the comparisons and, as expected, resulted in lower cognate rates due to false cognates being removed.

2.3.4. Comparisons involving South African Sign Language

Nokwazi (2017) is the only traceable study that has compared the SASL lexicon with other sign languages. Nokwazi used the modified Swadesh list and compared signs from SASL to ASL, BSL, ISL, PGSS, and English signs. English signs were defined as signs that are either fingerspelled with the manual alphabet or initialised, i.e. signs where the handshape corresponds to the first letter of the English translation. FAMILY (see fig 2.4) is an example of an initialised sign because the handshape is the same as the manual alphabet letter ‘f’ which is the first letter of the English word ‘family’.

(26)

Figure 2.4 FAMILY in SASL

Nokwazi does not elaborate on why this distinction was made as initialised signs are an intrinsic part of SASL.

Nokwazi used data from four informants, selected on the basis of school attendance: two who attended Wittebome School for the Deaf, which was established by the Irish Dominican order in Cape Town, one who attended Hammanskraal School for the Deaf, which was also established by Irish Dominicans in 1962, and one who attended St Vincent school for the Deaf in Johannesburg (Nokwazi herself). It is not clear why these schools were chosen. The historical background of these schools makes it probable that a bias towards ISL would be found but Nokwazi does not mention this. Nokwazi sourced signs for ASL from www.lifeprint.com and www.handspeak.com, and BSL from http://bslsignbank.ucl.ac.uk. ISL data was provided by an Irish SL signer; it is uncertain if this person is a native signer.

Nokwazi sent the list of lemmas in English to her informants and seemingly had them record the signs by themselves. It would not be possible to say for certain that the signs were elicited without another person’s involvement. It is also unclear if the informants received any instructions on their signing, i.e. should they use their everyday signs or attempt to recall the signs from the school they attended.

Nokwazi followed the McKee and Kennedy (2000) protocol where signs that differed in two or more phonological parameters were categorised as different. Nokwazi does not elaborate on her description of phonological parameters, e.g., whether the location of a sign recorded where the sign starts or where it ends or whether this is addressed in the movement parameter.

The similarity that the Wittebome signers have with ISL is explained through the school’s history with Irish Dominicans. However, one of the signers was 78 years old and had moved to Johannesburg after school. It may therefore not be accurate to still equate the signer’s signs as representative of Wittebome. This is supported by the fact that the two signers from Wittebome had different signs for WHERE.

While Nokwazi (2017) does mention iconicity as a limitation of her study, she fails to mention the influence it may have had on the findings. Table 2.4 below presents the overall findings of Nokwazi (2017), while table 2.5 presents the findings according to province (the total number of lemmas and signs elicited is not clear).

(27)

Identical signs Similar signs Total ISL 14 11 25 BSL 19 9 28 ASL 11 5 16 PG 3 0.3 3.3 English (fingerspelling/initialised) 7 1 8 80.3 Table 2.4 Nokwazi (2017) findings – Percentage of similar and identical signs

The percentages indicate the borrowings from the various languages identified in the first column. A total of 80.3% is given as borrowings from other sign languages; this is the total of identical and similar signs. It is not clear what Nokwazi did in instances where there was overlap with multiple sign languages e.g. one lemma may have a sign that is similar to both BSL and ISL and should therefore not be counted twice. The wordlist consisted of 100 words; it is not explained how a percentage of 0.3 was determined for similar signs in PGSS.

(28)

Western Cape Gauteng ISL 14% 18% BSL 14% 27% ASL 4% 15% PG 1% 4% Initialised 5% 6% Total Borrowing 38% 70%

Table 2.5 Nokwazi (2017) findings – percentage of borrowings per province

In table 2.5 the Gauteng column combines the results using data from St Vincent and Hammanskraal schools for the deaf and the Western Cape column shows the results comparing data from Wittebome school for the deaf. Once again, any overlap is not indicated, and totals are simply added up. The Western Cape has five schools for the deaf and Gauteng has seven schools for the deaf; one cannot attribute the variety used at one school as representative of the entire province.

2.4. Conclusion

Comparing the lexicons of sign languages has frequently been done in order to determine if the languages have a historical relationship. Woodward (1978, 1991, 1993, 1996) compared signs from various sign language varieties, yet provided very little information on methodology. Johnston (2003) stated very clearly the importance of determining the phonological significance of a sign’s features and suggests disregarding handedness. It is not known if Woodward included handedness in the comparisons nor if handedness was evaluated for phonological significance. The current study subscribes to Johnston’s methodology to a larger degree than to that of McKee and Kennedy (2000). Both of these studies included lemmas in addition to the Swadesh list and Johnston included variation as it was recorded in dictionaries.

Su and Tai (2009) raised the importance of iconic motivations within signs. If a study’s purpose is to determine the historical relatedness between sign languages one must pay specific attention to signs that are highly iconic. The current study identifies signs that are highly iconic, but only excludes signs if multiple sign languages have the same sign; the sign may therefore be highly iconic but still related to a sign from another language. The question on how to

(29)

determine if a sign is iconic is not clearly answered by Su and Tai. Al-fityani and Padden (2008) found higher than expected similarities between two sign languages that appeared to be unrelated and attributed this to iconicity. Su and Tai had a very clear focus on identifying iconic signs while Al-fityani and Padden makes mention of iconicity influencing the results. As the current study aims to identify loan signs, it will be necessary to provide details on the influence of iconicity on the results.

It is vital that data be gathered from reliable sources that are accurate representations of the sign language studied. Nokwazi (2017) elicited data from informants who were obviously no longer representative of the language variant. Various studies discussed here used dictionary entries as data sources, yet Johnston (2003) argued against this as the citation form of a sign may have variations in different settings.

Parkhurst and Parkhurst (2003) suggest a methodology that is more qualitative than quantitative. They emphasised that cognates can be signs that differ significantly in form. However, the current study is not attempting to identify cognates. While identical signs may indeed be cognates, Parkhurst and Parkhurst elaborated on a process of comparing historical and regional signs to identify signs that underwent phonological changes but had the same or similar roots. The current study uses signs used by college students who had finished school within the past several years, and who identified the signs as those used at their respective schools.

2.5. Research Questions

Given the preceding discussion, and the aim of the research outlined in chapter one, the research questions for this thesis are:

1. What is the extent of lexical variation in SASL? 2. What is the evidence for lexical borrowing in SASL?

(30)

Chapter 3 Methodology

In this chapter, the methodology of this study will be described. An approach was followed that was similar to Johnston (2003), as discussed in the previous chapter.

In section 3.1, the informants from which data were gathered are described. Section 3.2 provides detailed information on the materials used and elicitation, and the procedure is described in section 3.3. Finally, in section 3.4, the process of analysis is explained, including a description of SASL phonology.

3.1. Informants in the dictionary project

The National Institute for the Deaf (NID), South Africa, is continuously developing an on-line bilingual dictionary of SASL and English (www.nid.org.za/dictionary). NID has a training college on its campus and provides training and accommodation to between 120 and 140 deaf students annually. These students are the source from which data are being gathered to fill the dictionary.

Data available for the current study were taken from the data of NID dictionary project. These data were collected from 50 students who attended the NID College and who had previously attended a school for the deaf (see Table 3.1). The informants in the project needed to have attended the school for the majority of their schooling career to ensure that their lexicon was representative of the specific language variant used at their school. Three rounds of recording were done, but not all informants could attend each session due to scheduling conflicts. This resulted in less data from schools that those informants attended.

Initially informants were limited to one per school and were invited to the studio where recordings were done. By round three the recordings were being done at the College making it possible for more students to participate.

Informants represented a total of 20 schools for the deaf across five provinces in South Africa, Western Cape, Gauteng, Free State, Gauteng, and KZN. According to the Department of Basic Education, there are 43 schools for the deaf in South Africa; five of these are units at mainstream schools and several are schools for the deaf and blind. Four provinces were not represented: Northern Cape, Mpumalanga, Limpopo, and North West, which meant that the other 23 schools for the deaf were not included in this comparative study. The NID did not have students from these schools. In total 50 informants took part in the study with a mean age of 22;3. There were 27 males and 23 females.

(31)

Round Age School attended Gender Province

1 22 Efata M Eastern Cape

3 23; 22 Efata F; F Eastern Cape

1 20 Reuben Birin M Eastern Cape

2 20 Reuben Birin M Eastern Cape

1 22 St Thomas F Eastern Cape

2 22 St Thomas M Eastern Cape

1 26 Bartimea M Free State

2 26 Bartimea M Free State

1 27 Thiboloha F Free State

2 22 Thiboloha F Free State

1 20 Dominican Hammanskraal M Gauteng

2 20 Dominican Hammanskraal M Gauteng

1 24 Filadelfia F Gauteng 1 25 Filadelfia M Gauteng 1 18 MC Kharbai M Gauteng 3 22; 19 MC Kharbai F; F Gauteng 1 23 Sizwile M Gauteng 2 23 Sizwile F Gauteng 1 21 St Vincent M Gauteng 3 25; 24 St Vincent M; M Gauteng 1 22 Transoranje F Gauteng 3 22 Transoranje F Gauteng 1 20 Fulton F KZN 1 18 Fulton F KZN 1 24 Kwathintwa M KZN 2 23 Kwathintwa F KZN 3 22; 23 Kwathintwa M; M KZN 1 22 St Martin de Porres M KZN 2 22 St Martin de Porres M KZN 3 22 St Martin de Porres F KZN 1 26 Vuleka M KZN 2 23 Vuleka M KZN 3 28; 22 Vuleka F; M KZN 3 22; 21 VN Naik F; F KZN

1 20 De la Bat M Western Cape

2 20 De la Bat F Western Cape

3 28 De la Bat M Western Cape

1 21 Dominican Wittebome M Western Cape

2 20 Dominican Wittebome M Western Cape

3 23; 20 Dominican Wittebome M; F Western Cape

(32)

3.2. Data elicitation materials

In most lexicographical comparison studies, as discussed in 2.3, the Woodward (1978) list of lemmas is used. Multiple studies rely on dictionary data, which is very efficient, but as noted in 2.3.2 there are nuances that go unnoticed when only referring to the citation form of a sign (Johnston, 2003).

Many studies have included additional lemmas (McKee and Kennedy, 2000; Johnston, 2003; Al-fityani and Padden, 2008; Clark, 2017) because a larger lexical sample of the source language would result in a more accurate comparison. While the modified Swadesh list of lemmas contain lemmas that are core to the human experience, this study follows these researchers and added additional basic lemmas. This study was limited to the data available from the NID dictionary project (see 3.1). The lemmas used in this study’s comparison are listed in Table 3.2. In total 173 lemmas were used to elicit signs for the dictionary project. Lemmas from the Woodward (1978) list for which data were not available are: dance, grass, hunt, louse, snow, sun, with, and worm.

Round one of data elicitation for the dictionary project consisted of 47 lemmas elicited from informants from 19 schools for the deaf. Round 2 consisted of 53 different lemmas elicited from 14 schools for the deaf. Round 3 consisted of a further 73 lemmas elicited from 12 schools for the deaf, elicited signs are labelled as tokens. Dependent on the number of schools for each round, the potential number of tokens could either be 19, 14, or 12 for each lemma. Round one potentially has 893 tokens, round two 742 tokens, and round three 876 tokens. The missing data will be addressed in section 4.1.1.

(33)

ROUND 1 (47 lemmas – potentially 19 tokens per lemma)

Apple Ask Autumn Believe Bird* Blue

Bread Butter Cheese Child* Decide Dog*

Father* Fish* Green* Happen Help Like

Live* Lunch Make March Meat* Milk

Monday Month Mother* Parents Play* Potato

Promise Rain* Red* Sit* Snake* Start

Sunday Wait Warm* Week Win Wind*

Winter Work* Worry Year* Yesterday

Answer Aunt Black* Borrow Boss Breakfast

Bring Brother* Brown Buy Cat* Chicken

Cow Daughter Day* Dinner Elephant Find

Finish Follow Forget Grandfather Grandmother Horse

Hour Learn Lion Lose Minute Mouse

Night* Orange Pizza Porridge Purple Remember

Rice Salad See Sell Sister* Son

Stand* Summer Tell Think Travel Try

Uncle Understand Use White* Yellow*

All* Animal* Bad* Because* Blood* Count*

Die* Dirty* Dry* Dull

(boring)*

Dust* Earth*

Egg* Feather* Fire* Flower* Full* Good*

Grease (oil)* Heavy* How* Husband* Ice* If*

Kill* Laugh* Leaf* Lie* Long* Man*

Moon Mountain Name Narrow* New* Not*

Nothing Old* Other* Person* Pig* Right

(correct)*

River* Rope* Safe Salt* School Sea*

Search Sharp* Short* Sing* Smooth* Star*

Stone* Tail* Test Thin* Tree* True

Vomit* Water* Wet* What* When* Where*

Who* Why Wide* Wife* Woman* Wood*

Young

Table 3.2 Wordlist used in this study (*Lemmas that were part of Woodward (2003))

(34)

discussed with the facilitator the different signs they have encountered and assured the facilitator that they recalled their school’s signs very well. However, on several occasions, students could only recall their school’s sign after having a discussion with a peer. The reason for this is not clear but may be due to lack of use. During round 3 there was more than one informant representing a school and it was found that it was markedly easier for the students to recall their school’s sign when accompanied by a peer.

English was chosen as the elicitation language as it is the language of literacy used at NID. It was considered impractical to use multiple written languages on the flashcards due to the space it would take up. The Deaf facilitator held up flashcards that showed an English word and an image. It was a challenge to find appropriate images for many words, e.g., if, name, and wide. See Figure 3.1 for examples.

Figure 3.1 Examples of images used with lemmas

Informants were additionally asked to sign a sentence to illustrate that the sign was a semantic fit to the flashcard. In cases where the informant did not understand the word, the facilitator would explain the meaning while being mindful of possible influence from their own variety of SASL. In some cases, the meaning could not be conveyed, and no sign could be elicited. Several informants referred to the image first and provided incorrect signs; for example, sharp had a knife as an image and elicited the sign for knife. The facilitator would intervene in these cases and elaborate.

In rare cases the informant indicated that his/her school had no sign for the lemma. It cannot however be assumed that there is no sign for that lemma in that particular school. There was only one instance where multiple informants from a school indicated that they had no sign (for

(35)

sharp). In one instance one informant indicated they had no sign (for tree), but the other informant did.

The data were categorised per school for the deaf. The sign from each school was compared with each of the other schools following the analysis to be detailed in 3.4. This resulted in several sign types for each lemma; these sign types were then compared with the potential lexifier languages.

3.4. Analysis

In this section the process of analysis will be explained. A significant part of the analysis was to determine the number of types, i.e. how many different variants each lemma had. To determine this, the phonological features of each lexical item needed to be described prior to a comparison being done (3.4.1). The phonological features allow for a comparison to be done within the SASL lexicon to determine identical and similar signs. Iconicity is then addressed (3.4.2) where the criteria to exclude certain lemmas due to having a high degree of iconicity is discussed. Finally, the comparison to potential lexifier languages6_{is discussed (3.4.3).}

3.4.1. Phonological Description

The phonology of signs is commonly described in terms of four manual parameters: handshape, orientation, location, and movement; each of these will be discussed; non-manual features have a more grammatical function and is not included in a phonological comparison. Lucas and Bayley (2005) propose that there is more to sign language phonology than merely these four parameters. They suggest that the order in which phonological features are produced is indeed phonologically significant. Liddel and Johnson (1989) recognised that there are sequential elements within signs but claimed that the sequentiality is not phonologically important. Due to a lack of a model that sufficiently describes the phonological features of the sequential organisation within signs, this study will refer to the aforementioned four parameters.

(36)

3.4.1.1. Handshape

Handshapes were identified using a special font7 that showed the actual handshape image (see Figure 3.2). For one-handed signs, only the right hand was listed. Informants who were left-handed were transposed to right-handed to avoid possible confusion. Two-handed signs were listed with each hand’s handshape identified. The data revealed 52 different SASL handshapes.

Figure 3.2 Example of handshapes in the font library

The handshape of each sign was identified based on the initial shape of the hand. Many signs have handshape changes that occur sequentially.

Figure 3.3 Initial and final handshape of UNDERSTAND

(37)

Figure 3.3 illustrates the change in handshape for the sign UNDERSTAND. The initial handshape was documented as 6 and the final handshape as B. This change in handshape was documented as part of the movement parameter, which will be described in more detail in section 3.4.1.4.

Close inspection reveals that the initial handshape is not exactly the same as the documented handshape 6; the knuckle of the index finger is slightly extended whereas the documented handshape is a fist. In this case it had to be determined whether this specific handshape was phonologically or phonetically different; does the handshape with the knuckle slightly extended carry meaning? To determine this, a minimal pair needed to be found i.e. two signs needed to be found where the only difference was the handshape. A minimal pair could not be found, and this handshape was therefore deemed to be an allophone, the handshape difference does not carry semantic weight, for the documented handshape.

It is commonly known that minimal pairs in sign languages are a rare occurrence. Figure 3.4 demonstrates the purpose of finding minimal pairs. The two handshapes used in the two images are semantically significant because the location, movement, and orientation of the two signs are same with only the handshapes being different. This process was followed for each of the parameters where there was uncertainty on the phonological validity.

(38)

3.4.1.2. Location

The location of each sign refers to the area in space or on the body where the sign is performed. There are very specific locations that carry meaning. There are four primary locations where signs are articulated, the head, the upper body, the non-dominant hand, and the neutral space (Baker et al., 2016). These four areas were further divided into phonologically significant locations in the analysis. The initial location of the sign was identified. The top of the shoulder is one such location. Minor variations in the exact location on top of the shoulder do not carry meaning. If a sign moved to another location, it was identified as part of the movement parameter, but only the initial location was documented.

Figure 3.5 GREY

GREY, as shown in Figure 3.5, was articulated in the primary location of the head and the specific location was documented as the temple. Both of the signs in Figure 3.3 above were documented with the location as the chin.

3.4.1.3. Orientation

Orientation refers to the orientation of the palm and fingers. In most of the research reviewed as part of this study, only the orientation of the palm was considered in the comparison, if there was indeed any mention of the parameter details at all. The orientation of the fingers is a phonological part of the sign and cannot be ignored (Baker et al., 2016). The palm orientation could be any of the following:

(39)

● Towards signer ● Away from signer ● Up

● Down ● Left ● Right

● Towards each other ● Away from each other

Signs that were oriented diagonally were assigned either the left or right property as these vary phonetically, rather than phonologically. This follows the view that orientation can be more accurately described as the relationship between a part of the hand and the location (Baker et al, 2016). The orientation of the fingers was identified by the direction of the straightened fingers; in a sign that had a closed hand the imagined direction of the straightened fingers was used.

3.4.1.4. Movement

The movement parameter was a complex parameter to describe. The other parameters, for the most part, have quite specific properties, and phonetic variation was simple to detect. The movement parameter however has more options, e.g., speed of the sign, path of the sign, variations in the path, movement of the fingers. The speed of the sign was determined, in this case, to be phonetic and not included in the comparison. The movement path of the sign was described in terms of the general direction. More descriptive words were used as there was too much variation in this parameter to have a pre-determined list. Movements of the wrist or hands that changed shape or orientation were given the movement property of “Hand Internal Movement”.

(40)

Figure 3.6 Initial position of DIE (left) and final position of DIE (right)

3.4.1.5. The process of determining similarity or identity between signs: a comparison of two APPLE variants

To better illustrate the process of determining similarity and differences between signs, the process using the phonological description of two variants of APPLE will be demonstrated.

(41)

The handshape was considered first, and the handshape font library was used to determine the relevant handshape, which in the instance of APPLE-1 in figure 3.7 is <. The location is ‘In Front Of Mouth’ and the palm’s orientation is ‘Towards Signer’. The path of the right hand is ‘Up’ with no hand internal movements or orientation changes and repeated twice.

Figure 3.8. APPLE-2

The second variant of APPLE in Figure 3.8, APPLE-2, is clearly different to the previous sign, but similarities are apparent. The handshape is 1 and the location is the chin. The orientation is towards the signer and the movement is a tap which means a movement is repeated twice with contact. Only one phonological feature is identical across these two variants, the palm orientation.

(42)

The phonological descriptions of each token of a lemma were compared with each other to determine identical, similar, and different signs. On the basis of common practice in the literature (e.g., McKee and Kennedy, 2000; Johnston, 2003) signs that were identical or similar, i.e. the signs differed in only one phonological parameter, were counted as a single type. This results in a number of sign types per lemma. Different signs are labelled as different types. For example, the lemma apple had three types and these types were used for the comparison to signs from the potential lexifier languages.

3.4.2. Iconicity

As discussed in Chapter 2.3.4, Su and Tai (2009) and Parkhurst and Parkhurst (2003), among others, noted that sign languages exhibit a high degree of iconicity, i.e. there is a resemblance between the linguistic form of a sign and its meaning (Taub, 2001). It is highly likely that a person familiar with Western cultures but unfamiliar with any sign language would still be able to sign SLEEP such that the form overlapped with the sign used in a sign language. It would, therefore, not be accurate to conclude that the signs SLEEP in BSL and SASL are identical due to historical influence.

Iconicity is a scale; signs can exhibit a certain degree of iconicity i.e. signs can be highly iconic, or the iconic motivation could only become clear once the meaning is understood (iconic, but non-transparent). This aspect will be further discussed in Chapter 4.

(43)

Figure 3.9 demonstrates the sign STAND which is highly iconic. The vast majority of sign languages use the same sign. For the purposes of this study a sign was considered highly iconic if many sign languages used the same sign. The resource www.spreadthesign.com was used. Many of the sign languages used in the resource are historically unrelated with no contact to allow for borrowing to occur. The reason for the sign languages using the same sign can therefore attributed to iconic motivation; the two extended fingers in the example STAND emulate the legs of a person standing.

3.4.3. Comparison to lexifier languages

Following criteria determined by McKee and Kennedy (2000), a sign type (see 3.4.1) was compared to the sign used in seven potential lexifier languages on the basis of the four parameters. If there were no differences, the sign was categorised as identical, if there was one difference the sign was categorised as similar, and if there were two or more differences the sign was categorised as different.

Sign language Source Variants indicated

American Sign Language (ASL) www.aslpro.com No

British Sign Language (BSL) www.bslsignbank.ucl.ac.uk Yes Dutch Sign Language (NGT) www.gebarencentrum.nl/gebaren/va

n-dale-ngt-uitgebreid/

Yes

Flemish Sign Language (VGT) gebaren.ugent.be/ Yes

German Sign Language (DGS) www.spreadthesign.com No

Irish Sign Language (ISL) Irish native signer (p.c.) No Paget Gorman Sign System

(PGSS)

www.pagetgorman.org/how-to-sign No Table 3.3 Possible lexifier sign languages and their sources

Lexifier language No. lemmas available (max. 173)

No. of sign variants available

American Sign Language (ASL) 173 174

British Sign Language (BSL) 173 397

A lexical comparison of South African sign language and potential lexifier languages

DECLARATION

ABSTRACT

OPSOMMING

ACKNOWLEDGEMENTS

TABLE OF CONTENTS

Chapter 1

Introduction

Chapter 2

Literature Review

Chapter 3

Methodology